Audience Management Platform
B2B, 0 to 1, eCommerce, Adtech
Launched an audience management platform to enhance audience targeting, leading to 62% account adoption rate and 68% attach rate of new campaigns that use a saved audience.
My Role
User Research
Product Designer
User Testing
Team
1 Product Designer (me)
1 Product Manager
5 Engineers
Duration
3 months
Problem
Audience targeting was overly rudimentary
Criteo Retail Media is an advertising platform that helps brands run digital campaigns at retailers’ sites. Audience targeting is a vital part of advertisers' campaigns. However, the audience targeting feature in Retail Media was too basic, lagging behind the industry and failing to satisfy the needs of users. We would like to improve both the functionality and experience of audience targeting.
Solution
Building audiences using real data and managing audiences at scale
The final design is new audience creation and management experience that is easier, more flexible and more scalable. With retailer data, users can target audiences in different stages of the conversion funnel, and view the audience size as they create an audience.
After an audience is created, users can attach it to multiple campaigns, and store it in the account for future use. A standalone audience management platform was introduced so that users can manage their audiences outside campaigns at scale.
Research
Status quo
I interviewed account managers to understand the current process and pain points. Audience targeting was one step within the campaign setup workflow. The function was very limited, only supporting add-to-cart audience targeting. Another insight was about repetitive work. Users often wanted to target the same audience with multiple campaigns, but in the current setup they had to configure the same audience for each campaign.
Competitive research
I also conducted competitive research to understand how other platforms approached audience targeting. An interesting takeaway was the use of audience forecasting. The real-time forecasting would tell users the audience size as they tweak different settings. Some of the forecasting not only told users the audience size, but also gave users a sense of whether the audience was too broad or narrow.
Understanding technology
Since we would like to leverage retailer data enhance audience targeting, I talked to the Data team to understand what data were available, and how they could be translated into audience targeting. I learned that an audience was defined by four elements: retailer, shopping behavior, targeting parameters, and a lookback window.
Opportunity
HMW separate audience setup from campaign setup?
Based on research, I realized that having audience creation living inside campaigns would be the best experience. As audience targeting became stronger, the setup also become lengthier and more complex. As a result, I proposed a standalone audience creation and management experience in the platform. Users would be able to create audiences outside campaigns, and reuse the same audience across multiple campaigns.
Design
Audience creation form
I started out listing all the inputs needed to create an audience, and then created wireframes to explore the information and interactions. From users' feedback, the initial design was not intuitive enough. I explored several versions to make it easier to understand, and streamline interaction.
2-column layout to show selected products at glance
Users needed to select specific products or categories to complete the audience setup. Initially I used tabs to separate search results and selected results, so that the UI wouldn't be too busy. However, users complained that tab layout was inconvenient, as they had to switch between tabs to see the selected products. Base on this feedback, I changed the design to a two-column layout to improved visibility and reduce clicks.
Tree structure for intuitive category selection
Displaying product categories posed a challenge. Retailers often have deep category paths, which can be difficult to display in a user-friendly way. I proposed a tree structure, and worked with engineers to confirm the feasibility. But I ran into another challenge: the category could be as deep as 6 or 7 levels, making it hard to show in limited space. After talking with the PM, we decided to show only the first 4 levels of the category, which covered most use cases. Any level deeper would be too specific and result in very a small audience size.
Real-time audience size forecasting
Real-time audience size forecasting was crucial for users to understand the scale of their audience as they build it. Apart from displaying the audience size, I also designed a gauge to visualize the audience as narrow, balanced or broad to provide more guidance for users.
Audience table for easy management
After an audience is created, it will be stored in the account and displayed in the audience table. From engineers' feedback, the audiences would refresh regularly to include the latest data. To reflect this, I added a timestamp above the table to indicate data freshness. Since it takes some time for the system to match all the users after an audience is created, I added a ‘Populating’ status for the audience, so users would know when the audience was still being processed.
Outcome
62%
account adoption rate
68%
Attach rate of new campaigns that use a saved audience